## Magic Find and its efficiency: A statistical insight

• Quote from GNA

If you could tell me how many legendaries you found in how many hours played, you would help be a great deal with this thesis already.

Hektus is currently working with such an analysis (I presume he's still updating it) in this thread.

Working with statistics about legendaries is troublesome simply due to their low droprate. It creates a requirement for a huge sample size to even be in a position of which only a few thoughts can be postulated. The fact that we cannot assist our data collection with add-ons or programs makes such data collection even more troublesome and time consuming.

Edit: typo
• Well considering that majority of my legendarys were found while leveling my characters, only 3 of my 9 have been found at lvl 60, 2 staff and a 2h

Edit: Put 6 insted of 9 D=
• Nice work man, interesting read. As others said, this is how we knew it was, but still it's very interesting to see it demonstrated. This will hopefully remove some doubts people have about hidden agendas and whatnot.

By the way, as far as I can see your English is flawless, so don't worry about it
• Awsome work, thank you a lot for clearing those questions.

A small thought for all those poor Goblins we relentlessly decimate every day.
• I think it is important to note that magic find appears to not work on magic (blue) items.
From your experience with 0 mf and 50+ chance at blues at 100% mf we should reasonable see no white items
It appears from the data that mf is affecting rare drop more then blue drops.

However any calculation on mf could get skewed by guarranted drops, I'm not sure for sure-sure but I think gobos drop probably 2 blue + items and 2-4 white + items and then mf ups those drops. So while it will always increase rare chances those rare chances eat into the 2 guaranteed blues.

Note the reduction in white drops that probably is where we can actually figure out what is going on

"MF=0 : W=32.49%, M=57.67%, R=9.84%, S=0.00%, L=0.00% (Sample size: 437 items)
MF=76 : W=31.52%, M=54.57%, R=13.91%, S=0.00%, L=0.00% (Sample size: 460 items)
MF=150 : W=30.30%, M=51.30%, R=18.40%, S=0.00%, L=0.00% (Sample size: 462 items)
MF=234 : W=26.28%, M=44.54%, R=29.06%, S=0.00%, L=0.11% (Sample size: 898 items)
MF=252 : W=24.10%, M=42.57%, R=33.00%, S=0.00%, L=0.34% (Sample size: 888 items)"
It looks white drops went down by ~8% with 250 mf

Ideally we would want whites to drop off the table at that point you would start hitting diminishing returns as your only increasing rare/legend drops vs magic / rare/ legend

Personally with high mf I have gotten the most legendaries from bosses as they drop a lot of blue+ with high mf and NV stacks.

Take a look at the 0 mf data set and see what your lowest drops are (how many blue and how much white)
alot of times that rare comes from the blue column
• Very good post. I like the information.

So drop rates working with Magic Find work in the same way as combat tables in WoW. You might be familiar with those if you were ever a Warrior or Paladin tank. I knew as much already. Magic quality item drop chance decreases as Rare, Set, and Legendary (Or maybe Set and Legendary are grouped? Anyone know?) increases, because they "steal" the Magic quality drop percentages.

What I'm wondering is at what point do Set/Legendary drop rates start stealing from Rare drop percentages? I assume it's when Magic drop percentage hits 0%. Maybe that's an impossible number to hit. I dunno. Just something interesting to think about, and wonder how much Magic Find you would need to get exclusively Legendaries, or even to completely filter out Magic items.
• Quote from vastling

However any calculation on mf could get skewed by guarranted drops, I'm not sure for sure-sure but I think gobos drop probably 2 blue + items and 2-4 white + items and then mf ups those drops. So while it will always increase rare chances those rare chances eat into the 2 guaranteed blues.

Pretty much anything you kill (and still keep your sanity tracking it) will most likely have some sort of guarantied drop. I am not generalising things with my research: I am just providing a sample that can help people understand the mechanic behind MF.
In any case: it would be nice to know if goblins have a guarantied drop baseline. I think things evens out quite nicely, tbh, as I see many variations in drops, e.g. 1W+1M+1R, 3W+3R, 5M solely, I am also quite sure I had a seeker drop 3R and 3 gems but nothing else than that, etc.

Quote from aldrek

Very good post. I like the information.

...

What I'm wondering is at what point do Set/Legendary drop rates start stealing from Rare drop percentages? I assume it's when Magic drop percentage hits 0%. Maybe that's an impossible number to hit. I dunno. Just something interesting to think about, and wonder how much Magic Find you would need to get exclusively Legendaries, or even to completely filter out Magic items.

First of all: Thank you!
Second: The part of your post I have quoted pretty much summed up what I would like to get to. As it is right now there is a limit to MF. If I can somehow manage to assemble enough data points before things change too much it might be possible to make computations that can mimic the trends found for the data. These computations will involve parameters that are very interesting; especially what % MF is needed to wipe white drops off the board, or maybe even magic items.

- - -

And in general: thank you all for your feedback! I will come back in a few days to correct the original post to fit the inputs I am getting from you lot, however, I have a very important examn report to tend to now
• does the data for iLvl 63 items cover every item found, whites blues and rares or just rares?
• Quote from soniix

does the data for iLvl 63 items cover every item found, whites blues and rares or just rares?

It covers every item found - iLvl is rolled separate to item quality.
• Quote from Cyeron

Quote from soniix

does the data for iLvl 63 items cover every item found, whites blues and rares or just rares?

It covers every item found - iLvl is rolled separate to item quality.

ok thanks for this reply, I'm also farming these goblins for quite a time, i have 265% MF, got lucky with 2 Set Items but they were crap Tal Rasha's Guardianship 2 times ^^
• Just read that the drop rates are being changed in inferno.

source.

This will affect the appearence of the data in section 2.3 but the overall conclusion will stay the same.
Original post have been updated. Section 2.3 will be updated when a new sample size is available after the hotfixes.
• Cyeron, this was an excellent and most well thought out post. You have cleared up several misconceptions for me.

I'm a barbarian with about 150MF and have cleared Act 1 inferno a number of times working towards trying to survive Act 2.

When I first made a move towards MF, the difference was clearly evident. But I do have one request. Would you be willing to do some drop tests with full NV and going through elites, champs, goblins, and bosses? That is the case I think most people would be interested in now that you've cleared up all the mechanics without any NV.

For the record, I have seen two legendaries drop -- with max NV around 200 total and both were worthless junk. Both have been off random elites. I got exhalted plans for chestplate that dropped off the first group of zombies I ran into the beginning of the game when I didn't have any NV. Best 1H weapon I've seen so far is a 820dps mace that I'm using now. I've seen 3-4 1H weapons in the 700 range.
• Quote from marnes

When I first made a move towards MF, the difference was clearly evident. But I do have one request. Would you be willing to do some drop tests with full NV and going through elites, champs, goblins, and bosses? That is the case I think most people would be interested in now that you've cleared up all the mechanics without any NV.

I may do that at a later point, but there is a mechanic to NV that can contribute other than just the +MF it contributes: It will grant extra loot and guarantied 1 rare on a bosskill with 4 NV and 2 rares on a bosskill with 5 NV. So that would require me to get alot of data from kills with varying not only MF but also the amounts of NV buffs applied. It is a much larger work and honestly if I wanna pull that through in a proper time frame I may need help from the community.

The original post is currently targeted purely towards the MF affix and while the baseline drop chances may vary, it would be cool to gain enough data to make estimations on "MF caps" so to speak. The gobos in act II kinda provided me for an easy way to farm alot in a reasonable timeframe - hence why I chose them.

I will, however, note your request.
• Whoa , man! Now , that's a lot of work!! Really appreaciating your effort on this matter, the last days myself were running act 2 goblins, both with full MF gear, some MF pieces and full DPS, and couldn't decide which tactic i should stick to, but with your "little" maths here, i'm much more confident of which path i'm taking from now on :D.

Thanks a lot for the job here, even if its something that can seem quite obvious on a first look, on paper, and with all the nerfs / shady hotfixes / conspiracy shit ( xD i had to use that word ) , and with how res chests and gobos worked and work now, at least it helped me a lot.

BTW , i need to ask you something on this matter : As i said before , i'm on the hunt for this lil' bastards when i get bored of farming acts, and when i farm them for , let's say around an hour ( Random number , this just happens after a while doing it ) , they seem to only drop blues to me, but the time before this, it's usually 1 to 3 rares a kill. Now that you have been killing these buddies for hundreds , what do you think about this?
- The Demon hunt never ends.
• Quote from Vayne

BTW , i need to ask you something on this matter : As i said before , i'm on the hunt for this lil' bastards when i get bored of farming acts, and when i farm them for , let's say around an hour ( Random number , this just happens after a while doing it ) , they seem to only drop blues to me, but the time before this, it's usually 1 to 3 rares a kill. Now that you have been killing these buddies for hundreds , what do you think about this?

RNG is RNG.

I know that blizz will try to fix the gear swapping issue but I highly doubt it has taken effect yet. I actually don't think they know which way they want to fix it so I don't think gear swapping can be your issue (I dunno if you do that, just wanted to point it out).

If you look through my datasheet you will also find many streaks of which I get 2W+3M, 1W+4M or even 5-6M only items.

While it is an interesting theory (I did notice it a few times of which I thought "Hmmmmmmmmm.....") - I do think it is really just random. With high MF gear (I am currently running with 270, trying to get higher but I am getting really close to the limit without NV) you will potentially get the rolls of which you see 4+ rares on a single drop.

And thank you for your comment!
• Yeah i agree about it being RNG... but still.... i've been doing a double goblin route since the first-second week of the game on those goblins ( The first days at least the alcarnus one , the one i noticed myself before everyone and their mother knew about the ancient path one). This usually leads to fast "double-goblins" in a run, and it's weird, because since i said in my last post, i tried WITH and WITHOUT MF (First weeks were with 0% MF mostly ofc), and yeah , as you said , with MF i see more "tons of yellows" strikes, BUT , this seems to happen WHEN a rare drops.

Let me explain this : when i ran with no MF , yellows were dropping 1x and 2x mostly , 3x rarely , and if i can remember well, never x4 with no MF at all. Which as u explained here , it ends on more yellows with more MF. BUT , the part with the "all blues" streaks , its nearly the same with and without MF , AND , something than changes usually when this happens , didn't u notice that aroud 4 or 5 books are dropped? When usually its 1 or 2 with the rares , but blues strikes seem to drop this pack of books with them usually.

The weird thing here is , when i seem to reach this point, if i find 2 goblins in that game , its always like this for both of them in that game, both dropping tons or blues , or the pack with ~5 books + blues.

I always thought that there was some pattern for this to happen , or some diminishing return involved (At least i always had this feelling , and i also ran tons of gob runs) , since everyday i feel i got to a point where i hit a "Only blues" wall , and i have to stop the farm after a long time like this. Maybe its just me... but oh boy , it makes me wonder every single time xD

Sorry for the offtopic here , i don't want to drag the attention from your great post , but i needed in my inside to talk about this matter
- The Demon hunt never ends.
• Quote from Vayne

the part with the "all blues" streaks , its nearly the same with and without MF , AND , something than changes usually when this happens , didn't u notice that aroud 4 or 5 books are dropped? When usually its 1 or 2 with the rares , but blues strikes seem to drop this pack of books with them usually.

I am trying to understand what you mean so please correct me if I am misunderstanding your question :b

I did notice that tomes dropped along with the "blue streaks" as well but I also found tomes dropping in large amounts even when I got like 1W+2R or similar. One thing I found in my previous analyses is that bandits drop many more tomes than the rest of the types: maybe that's what you're seeing? There is a link in the original post, top part, of my previous analysis.

Someone did mention, however, that goblins most likely have a guarantied drop system. While I don't know of it, imagine the following imaginary example: if a goblin has a baseline dropchance of 1W + 2M (and additional loot can be added to this number, but those are certain on every kill), then with low MF you will on a lucky roll see 3+ of M items or 2+ M and 1R, and with high MF you will on an unlucky roll see the same. That could explain the observation, however, I do not know of baseline droprates for gobos.

Again, if you look through my datasheet you will see that there are "blue streaks" followed by some more or less "yellow streaks". I would like to point out that every farm run I took, 100 goblins were killed with no stopping. Since my data are continueous - and what you're thinking is that farming them for longer times will make more blues drop - I can personally say that I believe the blue streaks are not coming as a result of repetitive farming.
• It wouldn't be impossible for goblins to have increased returns on MF and Valor (the later would stop pure goblin rushes). With 5 valor stacks, I'm at 260% mf and I do everything in my power to kill those little shits, pretty much every time 4 rares drop (rarely drops to 3).
• Quote from Nivrax

It wouldn't be impossible for goblins to have increased returns on MF and Valor (the later would stop pure goblin rushes). With 5 valor stacks, I'm at 260% mf and I do everything in my power to kill those little shits, pretty much every time 4 rares drop (rarely drops to 3).

That is why I am not running with NV (at least not yet). I have a hunch that the guarantied rare drop from 5 stacks also applies to goblins.
• Quote from Cyeron

Quote from Vayne

the part with the "all blues" streaks , its nearly the same with and without MF , AND , something than changes usually when this happens , didn't u notice that aroud 4 or 5 books are dropped? When usually its 1 or 2 with the rares , but blues strikes seem to drop this pack of books with them usually.

...

Hmm maybe you're totally right about the bandits, after a while i stopped watching what i was killing in favor of what i was looting =P